Introduction

In Boston’s Dorchester, a mixed-income and culturally diverse neighborhood, a community group was fed up with the city government’s misperceptions of the neighborhood. They wanted to represent themselves and what was happening in their own backyards. So they built an artificial intelligence (AI) tool to make it happen.

The result, On the Porch, highlights the opportunity before all of us to change the way the public sector serves our communities—to rethink the public sector enterprise.

The problem is that the public sector may not know this work in communities is happening. It’s been focused on building its own capacity. Just in the first half of 2025, attention to AI has dramatically increased: In the United States, 700 state bills have advanced, many states are requiring on-the-job tech training and data literacy for all staff, and city-level pilots have mushroomed across the country.¹ There is indeed a lot happening, but mostly it’s focused on improving what government is already doing, not changing what it does.

AI is certainly billed as boosting efficiency, but also as a revolutionary technology. We believe that while the field has been focused on reforming, the real opportunity is in transforming.

We have some big ideas for how we can do it.

This work was conducted by RethinkAI: We are a coalition of researchers, activists, and practitioners long focused on the role of technology in the civic sector, brought together by the collective hunch that both the hype and the fear around AI are misplaced. Starting in the summer of 2023, we launched two parallel efforts to make sense of what is happening—an ongoing survey of trends and developments and a series of pilots with civic sector partners in three cities.

The ideas in this report were formed over the course of two years when we interviewed 40 officials; ran pilots in Boston, New York City, and San Jose; and reviewed the state of play throughout the country. We conducted this work with two goals: (1) provide a sober evaluation of the current landscape and a clear understanding of what AI implementation looks like on the ground, and (2) propose an alternative, results-based framework for civic and government actors to incorporate AI into their work.

To harness AI effectively, civic sector leaders need to build on reform efforts of the past. To genuinely rebuild trust in our democracy, we need to put results and people first. On the city and state levels, we have an obligation to protect our institutions, and an opportunity to reshape them to better serve residents by being more adaptable and responsive to community needs.

Our current moment is not entirely new. We have ridden tech waves in the past with varying results. Take the civic tech movement, of which we consider ourselves a part. This broad-based coalition includes a combination of practitioners from inside and outside of government. Over nearly two decades, civic technologists have launched thousands of projects, created and supported government apps, written books and white papers, and ultimately, created a community of practice committed to making public services better. This big tent of civic tech has adopted a kaleidoscope of different terms and names, including reinventing government, e-gov, smart cities, public interest technology, and on and on. Similarly, open data advocates shined a light on making the activities of government more transparent to the public, shifting attention from the mere delivery of services to the visibility and accountability of the institutions behind them.

Past efforts have led to significant change, but often change centered on improving process, leaving genuine transformation for another day. Rather than just making services better, faster, and cheaper, we should have focused more directly on what residents needed. We often fell into an efficiency trap. For those of us who have worked in public service, we found that bold ideas were met with a centrifugal force towards incrementally improving existing programs. Past choices lock institutions into familiar patterns, making it hard to break away from incremental fixes. The public administration term for this is “path dependency,” which essentially means that within governmental bureaucracies, it is almost impossible to break free from traditional policies and programs and embark on new paths. For someone who just lost their job and needs to feed their family, a simple, expedited enrollment process to their state’s food assistance program is critical. But over the long run, what that individual needs is access to resources to turn their employment and economic conditions around. We can’t just stop at access to public benefits.

Civic tech, despite good intentions, sometimes reinforced these paths instead of disrupting them. We weren’t doing enough to focus on outcomes or reverse the course of residents’ steadily diminishing trust in democratic institutions. Armed with tech tools and methods, our field celebrated basic government improvements on the merits of saving time and money. And while many of these measures led to improvements, a lot of people—especially in underserved communities—remain unconvinced by declarations of success.

This marginally-better-with-technology approach has run its course. Our public institutions are under attack. And many of our attempts at reform didn’t fully address residents’ needs, discontent, or apathy. We are at an inflection point, and we need to rethink the role of civic technology as institutions change to meet the advent of AI and a new federal landscape.

What follows is a high-level analysis of policies, programs, and practices in local governments in the United States and their partnering organizations. We share insights about general trends and novel approaches. And we identify open doors that we might collectively be able to walk through to shape AI to meet the needs of our democracy, as opposed to shaping our democracy to meet the needs of AI.